Efforts to align cancer care with the best available medical evidence have largely focused on disseminating guidelines, enhancing physician-patient communication, and revising payment schema. Although these are critical strategies, it is possible that another major influence on physician behavior has been overlooked - what if the practice of a physician's peers was a major determinant of whether she or he adopts one new cancer intervention or abandons another? We propose a novel investigation of the impact of physicians and physician patient-sharing networks on the adoption and abandonment of cancer interventions. Physician networks are groups of physicians who work with one another either directly or indirectly, through shared patients. New cancer management approaches could diffuse through such networks via social contagion, whereby ideas and behaviors spread through interconnected groups of people. It is particularly timely to explore new constructs that shape the delivery of cancer care, as the rapid diffusion of expensive, unproven, and sometimes ineffective cancer interventions poses a significant threat to the sustainability of the cancer car system and to the health of individual patients. For instance, the use of advanced imaging tests (AITs) such as PET scans and MRI for women with breast cancer has increased substantially over the past decade, despite little evidence supporting their use. Further, once an intervention has diffused into practice and become a standard of care, it can remain widely used even when there is evidence that it provides limited or no incremental benefit compared to less expensive or burdensome alternatives. For example, radiation oncologists have been reluctant to abandon the lengthier and more expensive standard courses of radiation for breast cancer, despite evidence that shorter courses yield similar outcomes. To address these concerns, we will focus on Medicare beneficiaries undergoing breast cancer care. We aim to (1) construct physician patient-sharing networks; (2) assess the impact of physicians and social contagion on the early adoption of AITs and the early abandonment of more costly and inconvenient longer-course radiotherapy; (3) identify physician and physician network characteristics associated with early adoption or abandonment of interventions; and (4) assess the population-level impact of physicians' adoption and abandonment of these interventions in terms of costs, care patterns, and outcomes, across physicians. Within any complex system, it is critical to understand factors that drive the behavior of individuals. A rigorous analysis of the complex web of physician and patient interactions in cancer care can not only provide a rich understanding of how physicians influence the behavior of their peers, but also identify a potentially powerful lever for inducing physicians to abandon ineffective interventions or discourage the use of unproven ones (Provocative Question E-3).